The tonoplast-associated citrate binding protein (CBP) of Hevea brasiliensis. Photoaffinity labeling, purification, and cloning of the corresponding gene.

A detailed comparison of citrate uptake into the vacuole-like lutoids of rubber tree (Hevea brasiliensis Muell. Arg.) and of malate and citrate transport into barley (Hordeum vulgare L.) vacuoles revealed very similar transport specificities. In order to identify proteins mediating the transport, two photoreactive analogues (N'-(2-hydroxy-5-azido)-diazo-N-3,5-benzenedicarboxylic acid and 5-azidoisophthalic acid) of malate/citrate were synthesized and found to efficiently inhibit citrate uptake into barley vacuoles (Ki = 18 microM) and Hevea lutoid vesicles (Ki = 27 microM). In vacuoles from both plant species, these photoaffinity probes specifically labeled a single protein with a molecular mass of 23.6 kDa. This citrate binding protein (CBP) was purified to homogeneity from Hevea lutoids, and amino acid sequences were determined for NH2-terminal and tryptic peptides. Using degenerate oligonucleotides of the NH2-terminal sequence, a cDNA coding for the CBP protein of Hevea was isolated. The cDNA codes for a precursor protein of 238 amino acids, containing an NH2-terminal 31-amino acid signal sequence for endoplasmic reticulum targeting, a prerequisite for vacuolar localization. The mature CBP does not show significant sequence similarities to any known primary protein structure and thus represents a member of a novel class of proteins.

A detailed comparison of citrate uptake into the vacuole-like lutoids of rubber tree (Hevea brasiliensis Muell. Arg.) and of malate and citrate transport into barley (Hordeum vulgare L.) vacuoles revealed very similar transport specificities. In order to identify proteins mediating the transport, two photoreactive analogues (N-(2-hydroxy-5-azido)-diazo-N-3,5-benzenedicarboxylic acid and 5-azidoisophthalic acid) of malate/citrate were synthesized and found to efficiently inhibit citrate uptake into barley vacuoles (K i ‫؍‬ 18 M) and Hevea lutoid vesicles (K i ‫؍‬ 27 M). In vacuoles from both plant species, these photoaffinity probes specifically labeled a single protein with a molecular mass of 23.6 kDa. This citrate binding protein (CBP) was purified to homogeneity from Hevea lutoids, and amino acid sequences were determined for NH 2 -terminal and tryptic peptides. Using degenerate oligonucleotides of the NH 2 -terminal sequence, a cDNA coding for the CBP protein of Hevea was isolated. The cDNA codes for a precursor protein of 238 amino acids, containing an NH 2 -terminal 31-amino acid signal sequence for endoplasmic reticulum targeting, a prerequisite for vacuolar localization. The mature CBP does not show significant sequence similarities to any known primary protein structure and thus represents a member of a novel class of proteins.
Malate and citrate play central roles in plant metabolism as intermediates of the TCA cycle and of CO 2 fixation, as components of pH homoeostasis and as the transport forms of reducing equivalents (for review, see Ref. 1). When accumulating at elevated levels in plants, these acids have been shown to reside in the vacuole. Oleski et al. (2) investigated the transport of citrate into tonoplast vesicles from tomato fruits and observed that citrate uptake was inhibited by malate and other dicarboxylic acids. Lutoids, which are vacuole-like vesicles in the latex of the rubber tree (Hevea brasiliensis), have been shown to accumulate citrate in vivo (3) (5.7 mM and 53 mM citrate in cytoplasm and lutoids, respectively). Citrate transport into lutoids in vitro was first described by d'Auzac and Lioret (4) using intact native lutoids, and later by Marin (5) using vesicles reconstituted from freeze-dried lutoids. Both native lutoids and tonoplast vesicles were shown to accumulate citrate from the outside against a concentration gradient. The uptake was stimulated by MgATP, and inhibitors of the lutoid tonoplast ATPase inhibited this activation of citrate uptake (5)(6)(7). Interestingly, the two dicarboxylates, malate and succinate, appear to cross the membrane by the same, or a closely related, transporter (8). In photosynthesizing protoplasts from barley leaves, malate and citrate are rapidly transferred to the vacuole (9). Transport of malate across the vacuolar membrane is not specific for the natural enantiomer, L-malate, since D-malate, citrate, and other di-and tricarboxylic acids behaved as competitive inhibitors (10 -12). Specific inhibition of the ATPase indicated that malate transport occurs by a secondary active transport mechanism (13). A detailed comparison of the uptake of citrate and malate into barley vacuoles showed that the characteristics of uptake are very similar and this suggested that malate and citrate are transported by the same transport system across barley tonoplasts (14). However, in order to determine the substrate specificity in detail and, furthermore, to understand the regulation of malate and citrate transport into vacuoles, the identification of the protein(s) mediating malate and citrate uptake is required. The malate/citrate carrier of barley tonoplasts (15) and the malate transporter of Kalanchoë daigremontiana vacuoles (16) have previously been functionally reconstituted and partially purified. The reconstituted carriers showed properties similar to those described for the carrier of intact vacuoles. However, further attempts to purify and identify the proteins mediating this transport activity were unsuccessful to date, due to limitations in the availability of tonoplast material and to the lack of methods for their detection.
In this study, we have compared citrate uptake into barley vacuoles and Hevea lutoid vesicles and were able to show that transport is efficiently inhibited by two photoreactive dicarboxylates. These photoaffinity probes specifically label a 23.6-kDa protein in both plant systems. We describe the isolation of the citrate binding protein of Hevea, the isolation of the corresponding cDNA and its gene, as well as the identification of homologous genes in barley.
Preparation of Lutoid Vesicles of Hevea-Latex was collected from trees of H. brasiliensis Muell. Arg., clone RRIM600, grown under natural conditions at the Rubber Research Center in Kuala Lumpur, Malaysia. Lutoids were sedimented by centrifugation at 4°C. The pellet (or bottom fraction) which constituted the crude lutoid fraction was frozen in dry ice/alcohol immediately after separation from the centrifuged latex and kept on dry ice until use. * This work was supported by the Swiss National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM (6)). The suspension was homogenized on ice using a glass homogenizer (1 g bottom fraction per 10 ml of medium) and vesicles were sedimented by centrifugation at 10,000 ϫ g for 15 min at 4°C. The pellet was resuspended and centrifuged twice under the same conditions. To avoid contamination of lutoid vesicles by soluble vacuolar compounds, additional homogenization and washing steps were performed, but these affected neither the protein composition visualized on silver-stained gels nor the labeling pattern. Finally, the sediment was resuspended in an equal volume of medium B (the same as medium A, but adjusted to pH 7.5 with Tris base) and used for tracer uptake experiments, for further purification, or for photoaffinity labeling experiments.
Uptake Experiments into Hevea Lutoid Vesicles-Uptake of [ 14 C]citrate and [ 14 C]sorbitol (both from Amersham, United Kingdom) was measured by a modification of the method described by Marin et al. (6). Vesicles were incubated at 25°C in medium B containing 3 kBq/200 l of [ 14 C]citrate or [ 14 C]sorbitol and solutes as indicated. At each time point four samples of 200 l were added to 1.8 ml of ice-cold medium B and centrifuged for 30 s at 14,000 ϫ g. The supernatant was removed and the pellet washed with an additional 900 l of medium B. The pellet was solubilized in 300 l of 10% SDS and radioactivity was determined by scintillation spectrometry.
Functional Reconstitution of the Citrate Carrier of Hevea Lutoids-Lutoid vesicles were solubilized with 2% Triton X-100 in 20 mM Tricine, pH 7.5, for 30 min on ice with vortexing every 5 min for 30 s. Insoluble material was sedimented (2 ϫ 20 min, 14,000 ϫ g) and the supernatant was filtered through a 0.2-m filter. Solubilized proteins were desalted on PD10 columns (Pharmacia) which had been equilibrated with 1% Triton X-100 in 20 mM Tricine (pH 7.5). Eluting fractions were diluted to a Triton X-100 concentration of 0.5% with an equal volume of double concentrated medium C (600 mM mannitol, 150 mM Mes, pH 5.5). Liposomes were generated in medium C (300 mM mannitol, 75 mM Mes, pH 5.5) as described (15) using 115 mg of phosphatidylcholine per 10 ml of medium C.
To energize the accumulation of citrate in proteoliposomes, a H ϩ gradient was generated by alkalinization of the medium. For this, one part of medium D (300 mM mannitol, 75 mM Tricine, pH 8.5, plus 10 l of 10 N KOH per 10 ml of medium) was added to two parts of proteoliposomes resulting in a proton gradient of approximately ⌬pH 2.8.
The assay medium for citrate uptake contained 2.2 kBq of [ 14 C]citrate per 100 l of assay medium (43.3 MBq/mmol) and solutes as indicated. After adjusting the concentration of organic acids to 10 mM (with citrate), proteoliposomes were separated from the incubation medium by a Dowex 1X8 -100 column (15).
Hevea lutoid vesicles were prepared as described for the uptake experiments. Concentration of the photoaffinity label was 15-20 M. Photolysis of the azido derivatives was achieved by illuminating the sample in microtiter plates, placed on ice, with UV light of max 366 nm (Camag; 420 milliwatts/cm 2 ). Illumination time was 5 min.
After fluorography gels were detached from filter paper by soaking in 4% glycerol and stained with silver nitrate (19).
Purification of the CBP of Hevea Lutoid Vesicles-Solubilized lutoid membrane proteins (see functional reconstitution) were separated on equilibrated (1% Triton X-100, 20 mM Tricine, pH 7.5) Sephadex G-75 columns (superfine, Pharmacia; bed volume 45 ml, 30 cm ϫ 1.5 cm). Proteins were eluted by hydrostatic pressure with a flow rate of 0.06 ml/min. Fractions of 0.6 ml were collected, and aliquots of 100 l were labeled. Fractions containing CBP were loaded on an equilibrated affinity column for the preparation of which 5-aminoisophthalic acid had been coupled to N-hydroxysuccinimide-activated agarose gel beads (Affi-Gel 15, Bio-Rad) according to the manufacturer's instructions. The column was washed with 3 volumes of 20 mM Tricine, pH 7.5, 1% Triton X-100, and proteins were eluted with a citrate gradient (0 -20 mM). For assays of transport activity and labeling experiments, citrate was removed by chromatography on PD10 columns (Pharmacia).
Sequence Determination of the NH 2 Terminus and Tryptic Peptides-Tryptic peptides of the purified protein were separated on an Aquapore RP300 HPLC column (100 ϫ 1 mm; Brownlee Tm column) with a linear gradient from 0.1% aqueous trifluoroacetic acid to 0.1% trifluoroacetic acid in 80% acetonitrile. Sequence analysis was performed with an Applied Biosystems 470A protein sequencer with a 120A analyzer.
Polymerase Chain Reaction (PCR)-Isolation of a cDNA fragment coding for an NH 2 -terminal fragment of HbCBP was achieved by amplifying inserts of a cDNA library of Hevea in gt10 using gt10 primers (BioLabs). An aliquot of the reaction mixture was used as template for a second PCR using primers Hb1 5Ј-CCNACNGAYGGNTTYAC corresponding to positions 125 to 141 (cDNA sequence in Fig. 6) and Hb2 5Ј-YTGDATNACRAARTTRTC corresponding to positions 178 to 161. DNA fragments of 50 -60 bp in size were isolated, subcloned into pBluescript SK(ϩ), and the sequence determined using T7 DNA polymerase (Pharmacia, Dü bendorf).
A fragment of the HbCBP gene was amplified using inverse PCR. Genomic DNA from young Hevea leaves (20) was digested with PstI, circularized with T4 DNA-ligase, and used as template for the primers Hb3 5Ј-CAGAGCAAAAATGTCTGTCC corresponding to position 119 to 100 (cDNA sequence in Fig. 6) and Hb4 5Ј-GATCCAACTGATGGGT-TCAC corresponding to positions 122 to 141. An aliquot was used as template for a second PCR using primers Hb5 5Ј-GCTCACAAGAAG-CAAAAGGG corresponding to position 91 to 72 and Hb6 5Ј-TGAGGT-GCCATTAACAGAGG corresponding to positions 142 to 161. The single amplified fragment of about 1.4 kb was isolated, subcloned, and sequenced.
Screening of cDNA Libraries-A Hevea latex cDNA library (kindly provided by Dr. S. Sivasubramaniam, National University of Singapore) was screened under stringent conditions using either the 54-bp PCR fragment coding for the NH 2 terminus of CBP or the 1.4-kb PCR fragment containing the 3Ј-region of the gene as 32 P-labeled probe. Positive clones were plaque purified, DNA was isolated, the insert fragments were subcloned into pBluescript SK(ϩ) and sequenced using synthetic primers. The DNA sequence has been submitted to the EMBL Data Bank under accession number X89855.
Southern Blot Analysis-Genomic DNA was isolated from young leaves of Hevea or barley according to Murray and Thompson (20). Electrophoretic separation and transfer to Hybond N membrane (Amersham) were done according to the manufacturer's instructions. Hybridization was performed under nonstringent conditions using the 630-bp COOH-terminal cDNA fragment as 32 P-labeled probe. Autoradiography was performed with a PhosphorImager (Molecular Dynamics).

Labeling of Barley
Tonoplasts-Transport of malate and citrate across barley tonoplasts has been characterized in detail using intact vacuoles and functionally reconstituted tonoplast proteins (10,14,15). Although protein fractions mediating malate transport were highly enriched by two purification steps, transport activity could not be confined to a single polypeptide (15). In a novel approach to identify putative carrier proteins for malate and citrate, photoreactive analogues of the two acids were synthesized. Since isophthalate inhibited citrate and malate uptake very efficiently, derivatives of this compound were used as photoaffinity labels. Indeed, HABDA and AIPA were found to be potent competitive inhibitors of citrate transport and therefore seemed to be suitable for effi-cient labeling (Fig. 1, Table I, A). The apparent K i of HABDA with respect to citrate uptake into barley vacuoles was 18 M.
Barley tonoplasts were isolated and the proteins subjected to chromatography on dry hydroxylapatite, after which malate and citrate transport can still be measured in a reconstituted system (15). This purified protein fraction was irradiated in the the presence of [ 3 H]AIPA and in the presence or absence of competing citrate. A specifically labeled protein with an apparent molecular mass of 27 kDa was identified after SDS-PAGE and fluorography (Fig. 2). Since binding of the photoaffinity probe could be prevented by addition of 10 mM citrate before irradiation, the labeled protein was a good candidate for a putative malate and citrate transporter. However, for purification of the labeled polypeptide and for more detailed analyses of binding of the photoaffinity probes, the availability of barley tonoplasts was limiting. Vacuoles from Hevea latex seemed to be an advantageous source of tonoplast material, since large amounts can be isolated with a minimum of preparative efforts. To demonstrate similarity to the described transport systems in barley mesophyll vacuoles, it was, however, necessary to characterize citrate transport specificity into lutoid vesicles in more detail.
Characterization of Citrate Transport into Hevea Lutoid Vesicles-Citrate transport into lutoids was stimulated by MgATP and the K m of 4.3 mM was close to the previously reported K m of 5-8 mM data (5), even though we used frozen lutoid fractions rather than freeze-dried material to generate lutoid vesicles. To test the substrate specificity of the citrate uptake system of lutoid vesicles, the rates of citrate transport in the presence of various carboxylates and other anions were determined. Of the naturally occurring dicarboxylates tested, only malate and succinate inhibited citrate transport, while oxaloacetate had no significant effect (Table I, A). The monocarboxylate ␣-hydroxybutyrate did not affect citrate uptake either. Isocitrate weakly inhibited citrate uptake, whereas neither sulfate nor glutamate were inhibiting. The photoaffinity probe HABDA competed for citrate uptake into lutoid vesicles with a K i of 27 M. As shown for barley vacuoles, pyridoxal phosphate was a strong inhibitor for citrate uptake in Hevea lutoids as well (not shown). To demonstrate specific uptake, [ 14 C]sorbitol was used as a control; uptake was slow and similar under all conditions (not shown; sorbitol: 1.5 nmol ϫ g of lutoids Ϫ1 ϫ min Ϫ1 ; citrate: 42.3 nmol ϫ g of lutoids Ϫ1 ϫ min Ϫ1 ). As the degree of inhibition of citrate uptake into lutoid vesicles was similar to that observed for citrate and malate transport into barley vacuoles (10,14) ( Table I, A), we conclude that the vacuolar citrate carriers of the two plant species have similar substrate specificities.
Labeling of Hevea Lutoid Vesicles-Hevea lutoid vesicles were irradiated in the presence of [ 3 H]HABDA with or without competing citrate. As in barley tonoplasts, a polypeptide with an apparent molecular mass of 27 kDa was labeled, and citrate competed with the binding of the photoaffinity label (Fig. 3). Labeling with [ 3 H]AIPA gave similar results (data not shown). Fluorography of control samples incubated for 5 min in the FIG. 1. Structure of HABDA (A) and AIPA (B). The position of the tritium label is marked by an asterisk.

FIG. 2. Photoaffinity labeling of barley tonoplast proteins partially purified by chromatography on hydroxylapatite with [ 3 H]AIPA.
Purified barley tonoplasts (corresponding to 1.55 ϫ 10 8 vacuoles) were irradiated for 5 min at 366 nm in the absence (Ϫ) or presence (ϩ) of 10 mM citrate and proteins subsequently separated by SDS-PAGE. In A the gel stained with silver nitrate is shown; in B the corresponding fluorogram is presented. The fluorogram was exposed for 6 months. The molecular masses of marker proteins are given in kDa.

Tonoplast-associated CBP of H. brasiliensis
dark or under the same dimmed light conditions as used for the uptake experiments, did not reveal any labeled protein (not shown). Separation of proteins on polyacrylamide gels without urea resulted in a slightly lower apparent molecular mass of the labeled protein (about 23 kDa). To find out whether the labeled polypeptide was an integral component of the lutoid membrane, loosely associated membrane proteins were removed by repeatedly freezing and washing the vesicles with 0.5 M KCl or by washing with 0.2 M Na 2 CO 3 (21). After KCl washes most of the labeled protein was retained in the membrane fraction. However, washes with Na 2 CO 3 released part of the CBP into the supernatant. But even after Na 2 CO 3 washing labeled CBP can still be detected in the membrane fraction (not shown), indicating that although the labeled protein is tightly associated with the lutoid membrane, the CBP is probably not an integral membrane protein.
To determine the specificity of binding of the photoaffinity probe, di-and tricarboxylates and other anions were tested for their ability to prevent binding of [ 3 H]HABDA to CBP (Table I,  B). Proteins were separated by SDS-PAGE, gels were subjected to fluorography, and the intensity of labeling was determined by densitometric measurements. Citrate, isocitrate, isophthalate, HABDA, and benzenetricarboxylate inhibited binding of the photoaffinity label very efficiently, whereas malate and succinate were less effective. Oxaloacetate, ␣-hydroxybutyrate, chloride, sulfate, and nitrate (not shown) at the same concentration did not have any significant effect. With the exception of isocitrate, which prevents efficient binding of the photoaffinty label but is only slightly inhibitory in the uptake experiments, inhibition of uptake agreed very well with protection from affinity labeling. To determine whether the labeled protein was able to mediate citrate transport, the CBP of Hevea was purified and a reconstitution system was established which allowed measurement of citrate transport activity.
Purification of the CBP of Hevea Lutoid Vesicles-Lutoid vesicles were solubilized in Triton X-100 and proteins were separated on Sephadex G-75 gel filtration columns. Fractions were tested for specific labeling by [ 3 H]HABDA. For further purification of CBP, an affinity column using aminoisophthalate as ligand was used and proteins bound to the column were eluted with increasing citrate concentrations. Elution with 4 to 8 mM citrate resulted in highly purified CPB (Fig. 4).
Reconstitution of Citrate Transport Activity of Lutoids-Total lutoid membrane proteins reconstituted into liposomes showed a time-dependent citrate uptake which could be inhibited by an excess of citrate, malate, isophthalate, and pyridoxal phosphate, whereas no inhibition was observed by sulfate or oxaloacetate (Table II). Dissipation of the proton gradient (which was employed to energize uptake) by the two protonophores FCCP and CCCP, or by sonication of the liposomes after generation of the proton gradient resulted in a drastically reduced citrate transport activity. In contrast to the reconstituted malate or citrate carrier from barley tonoplasts (15) for which an imposed membrane potential was used as driving force for malate uptake, no uptake of citrate could be measured under these conditions with reconstituted lutoid proteins. During purification of CBP, fractions were tested for their ability to mediate citrate transport after functional reconstitution. No

Tonoplast-associated CBP of H. brasiliensis
citrate transport activity could be detected after separation on a Sephadex G-75 gel filtration column. However, when solubilized proteins of lutoid vesicles were directly loaded on isophthalate-Sepharose affinity columns, proteins eluting at citrate concentrations of 4 -8 mM (this fraction contained the CBP) showed citrate transport properties similar to total lutoid membrane proteins when functionally reconstituted into liposomes (Table II). Citrate transport characteristics were as described for lutoid vesicles or total reconstituted lutoid membrane proteins, in that oxaloacetate did not reduce the rate of citrate uptake, whereas pyridoxal phosphate, citrate, and the two ionophores CCCP and FCCP caused significant reduction in citrate transport. However, additional purification steps resulted in very low uptake activities which did not allow us to determine specificities for the highly purified CBP. Isolation of a cDNA Coding for CBP of Hevea-The purified protein fraction (Fig. 4) was used to determine the amino acid sequence of the NH 2 terminus of HbCBP and of peptides generated by digestion with trypsin (Table III). Degenerate oligonucleotides from the NH 2 terminus of CBP were used in a PCR with latex-specific cDNA from Hevea as template. The resulting isolated 54-bp PCR product was used as a probe to isolate a 227-bp fragment encoding the NH 2 terminus of the purified CBP from a latex-specific cDNA library. Since screening of several latex-specific cDNA libraries with the 227-bp fragment as probe did not result in additional cDNA clones encoding the COOH terminus of CBP, the gene was isolated by inverse PCR using nested primers and PstI digested and subsequently circularized genomic DNA as template. This resulted in the isolation of a 1.4-kb genomic fragment. The 5Ј end of the genomic PCR fragment was identical to the 3Ј end of the 227-bp cDNA fragment (Fig. 5) and contained an open reading frame including sequences located 3Ј of the EcoRI site terminating the 227-bp NH 2 -terminal cDNA fragment. By using this genomic 1.4-kb PCR product as a probe to screen the cDNA library, a 630-bp cDNA fragment encoding the COOH-terminal moiety of CBP was isolated. As depicted in Fig. 5, the isolated NH 2 -and COOH-terminal cDNA fragments represent clones of the same gene, adjacent to the EcoRI site. The genomic 1.4-kb PCR fragment covers the entire cDNA sequence starting from primer HC6 at position 142 and contains only one intron of 174 bp at position 311. The first ATG of the assembled CBP-cDNA sequence resides at position 29 (Fig. 6). At position 857 the CBP-cDNA carries a poly(A) tail, with a putative polyadenylation signal 30 bp upstream. All sequences obtained of tryptic peptides were found in the deduced amino acid sequence ( Fig. 6 and Table III). HbCBP encodes a precursor protein of 238 amino acids with a 31 amino-terminal extension resembling signal sequences for endoplasmic reticulum targeting with a central hydrophobic core, positive charges in the first five amino acids (Met, Lys, Met, Lys, Arg) and small, neutral amino acids at the cleavage site at position Ϫ1 (Ala) and Ϫ3 (Cys) (Fig. 7A) (22, 23). The mature protein, starting at amino acid 32, contains 207 amino acids and has a calculated molecular mass of 23.6 kDa.
A data base search for related protein sequences revealed no significant similarities to known primary protein structures. Analysis of the hydrophobicity of the deduced amino acid sequence demonstrates that CBP is hydrophilic and contains only one hydrophobic stretch which might not be sufficient to act as a membrane spanning domain (Fig. 7B) (24). Predictions for the secondary structure indicate that no amphipathic helices large enough to span the lipid bilayer are likely to be present in CBP. The hydrophobic NH 2 terminus of the primary translation product cannot be used for integration into the membrane since the signal peptide is not present in the mature protein.
Homologous CBP Genes in Hevea and Barley-To determine the number of CBP encoding genes in Hevea, Southern blot analysis was performed under nonstringent conditions using the 630-bp COOH-terminal cDNA fragment as probe (Fig. 8A). Independently of the used restriction enzymes, only one or two bands are detectable on the blot. This simple pattern suggests that CBP is encoded by a single gene per haploid genome in Hevea. Interestingly, under comparable stringency the rubber  tree cDNA probe cross-hybridized with genomic DNA of barley (Fig. 8B). DISCUSSION The concentration of malate increases in C 3 plants during light-dependent nitrate reduction, and in plants exhibiting CAM metabolism malic acid accumulates as a store for CO 2 during the night. However, the concentration of malate in the cytosol remains rather constant and excess malate is transferred rapidly into the vacuole, where it accumulates to considerable levels (for review, see Ref. 1). Diurnal fluctuations of malate in C 3 and CAM plants thus predominantly represent changes in vacuolar malate concentration and require malate transport across the tonoplast. Besides malate, citrate levels are subject to substantial day-night changes in CAM plants (25). Vacuolar malate and citrate transport have been shown in a variety of species and have been characterized at the biochemical level in barley vacuoles and tonoplast vesicles of CAM plants (10,12). In barley vacuoles, citrate and malate appear to be transported by the same system (14). Purification of the proteins mediating the transport of malate and citrate has been limited due to the difficulties in obtaining tonoplasts from barley vacuoles in sufficient quantities. Hevea lutoids represent an attractive source for the purification of tonoplast proteins since they can efficiently be isolated from latex sap. Previous investigations on citrate uptake into lutoids focused mainly on the energetics of transport (26). The detailed comparison presented here has now revealed that citrate transport into Hevea lutoid vesicles exhibits characteristics similar to those of malate and citrate uptake into barley vacuoles.
Similarity of Biochemical Properties of Citrate Transport-Malate, citrate, isophthalate and HABDA proved to be inhibitors of citrate transport in barley tonoplasts and Hevea lutoid vesicles. In contrast, neither ␣-hydroxybutyrate, oxaloacetate, glutamate, nor sulfate caused marked inhibition of uptake. However, differences between the vacuolar citrate transport activities of the two plant species were found upon reconstitution into liposomes. Even though substrate specificity and K m did not differ between vacuoles and lutoid vesicles, the respective driving forces for uptake did differ. Malate (15) and citrate 2 uptake into proteoliposomes of solubilized and reconstituted barley tonoplast proteins was driven by a membrane potential and was strictly voltage-dependent. The exclusive role of ⌬ as driving force for malate uptake has also been demonstrated for the reconstituted malate carrier of K. daigremontiana vacuoles (16), and voltage dependent channels involved in vacuolar malate transport have been described for CAM plants (27,28). In contrast, uptake of citrate could not be measured unless a H ϩ -gradient was generated as driving force (inside acidic). This would in fact support the exchange of citrate for H ϩ as proposed by Marin (5). Another explanation would be that an acidic environment inside the lutoids is required for efficient dissociation of citrate from the carrier. The use of photoaffinity probes strongly inhibiting citrate uptake nevertheless allowed the identification of apparently similar proteins.
Specificity of Binding of the Photoaffinity Labels-Radiolabeled photoreactive dicarboxylates specifically labeled a protein with a molecular mass of 23.6 kDa both in barley tonoplasts and in Hevea lutoid vesicles. With the exception of isocitrate, substrate specificity for binding of the photoaffinity probe to lutoid vesicles in the presence of competing anions was similar to inhibition of citrate uptake into these vesicles as well as barley vacuoles. Even though it was not possible to demonstrate citrate transport activity for the highly purified CBP, the data strongly suggest that CBP is involved in vacuolar citrate transport in both plant species. However, neither a citrate transporter from Salmonella thyphimurium (29), the C4 dicarboxylate transporter from Rhizobium meliloti (30), nor the dicarboxylate transporter of the peribacteroid membrane of soybean nodules (31) could be labeled under the same conditions (not shown).
Sequence Analysis of CBP and Similarities with Other Proteins-Purification allowed the isolation of CBP and the subsequent cloning of the respective cDNA. A data base search for related sequences revealed that CBP does not share significant homology with any known protein sequence. HbCBP therefore 2 D. Rentsch and E. Martinoia, unpublished results. encodes a member of a new class of proteins. The deduced amino acid sequence contains a 31-amino acid signal peptide not present in the mature protein. This leader sequence is thought to be involved in targeting the protein to the endoplasmic reticulum based on its homology to endoplasmic reticulum targeting signals (22,23). In contrast, integral membrane proteins such as the tonoplast intrinsic protein and the plant plasma membrane permeases for amino acids and ammonium do not seem to contain cleavable signal peptides for endoplasmic reticulum targeting (32)(33)(34). Because of its isolation from lutoids, CBP is expected to contain additional information for vacuolar targeting.
Analysis of its hydrophobicity indicates that CBP does not represent an integral membrane protein with multiple hydrophobic domains. This is in agreement with biochemical data showing that CBP is loosely associated with the lutoid membrane. In a parallel case, an affinity probe for sucrose allowed the identification of a hydrophilic 62-kDa sucrose binding protein (35) which is associated with the plasma membrane and appears to be involved in sucrose transport. The low hydrophobicity of CBP indicates that it does not represent the actual pore-forming subunit of the putative citrate transporter. However, binding of the photoaffinity probe and inhibition characteristics of citrate transport suggest that CBP may represent a peripheral subunit involved in substrate recognition. Of the few tonoplast proteins identified to date, the vacuolar ATPase is an example of a heteromultimeric complex (36). In Escherichia coli, amtA probably represents a cytoplasmic component of the ammonium transporter (37). Periplasmic proteins responsible for binding and subsequent transport of substrates (38) are required for the uptake of various solutes and provide further examples for transport not being mediated by a single polypeptide.
Conclusion-Photoaffinity labeling allowed the identification of a new protein which may represent a subunit of a vacuolar malate and citrate transporter. Further biochemical experiments, as well as the isolation of homologous genes and analyses of the function in transgenic plants are required to unambiguously define the function of CBP.